DESIGN AND MODELING OF AN ELECTROSTRICTIVE INCHWORM ACTUATOR
A. Suleman1, S. Burns and D. Waechter2
University of Victoria
Department of Mechanical Engineering
Victoria, B.C., V8W 3P6, Canada
Abstract
Currently, there is a strong and increasing demand for electric actuators. Applications
include the automotive, aircraft and space industries. The paper presents a proof-of-
concept design of a large displacement and medium force inchworm actuator. The
technology considered uses an electrostrictive mechanism that “walks” inside an outer
casing. This motion emulates an inchworm, summing small steps to achieve large
displacements. The detailed design of the electrostrictive inchworm actuator was
performed using finite element analysis. A prototype was constructed and tested and the
performance matched well with the numerical simulation results.
1. INTRODUCTION
Over the past decade, new fields of engineering have emerged from a surge of invention
and innovation, led by multifunctional materials. In particular, the emerging science of
multifunctional materials has spurred progress in the engineering field to enhance the
performance of structural systems. The new technologies have invited us to revise the
engineering rules, not only because they spur new industries but also because they
1 IDMEC-IST, Principal Investigator, Corresponding author; Tel: 250-721 6039; Fax: 250-721 6051; E-mail: [email protected] 2 Sensor Technology Ltd, Collingwood, ON, Canada; Tel: 705-444 1440; Fax: 705-444 6787
1
embody a sweeping capacity to lower the weight and cost of designing and
manufacturing new structural systems while improving overall performance.
Electromagnetic linear actuators have existed for decades. This device delivers limited
stroke and force as the total energy is restricted by the electrical energy stored in the
coil. In addition, the device is rather bulky, expensive, limited in frequency (due to large
inertia) and presents a potential safety hazard in the case of a power failure.
There are several key technological developments that have combined to establish the
potential feasibility of electrical actuator systems. One of the advances is the
development of functional materials and their utilization in devices such as distributed
actuators and sensors. The other development comes from the field of electrical
engineering, with the advent of new algorithms and signal processing technologies. The
pursuit of advanced materials with multifunctional properties has opened up new
horizons in terms of actuation simplicity, compactness and miniaturization potential.
The most recognized types of materials are shape memory alloys, magnetostrictive
materials and piezoelectrics, which develop strains (or displacements) when exposed to
thermal, magnetic and electric fields, respectively.
The electrostrictive phenomenon is a nonlinear property which exists in all dielectric
materials. Some of these materials can offer higher electrically induced strains with
lower hysteresis than piezoelectric materials. The strain is proportional to the square of
the applied electric field and independent of its polarity. The most promising
electrostrictive material is lead magnesium niobate (PMN); however, this material is
still not widely available on the commercial market. Constitutive models for
electrostrictors are not as mature as models for piezoelectrics, due to the non-linearities.
2
Piezoelectric behavior can be manifested in two distinct ways. The direct piezoelectric
effect occurs when a piezoelectric material becomes electrically charged when subjected
to a mechanical stress. As a result, these devices can be used to detect strain, movement,
force, pressure, or vibration by developing appropriate electrical responses, as in the
case of force and acoustic sensors. The converse piezoelectric effect occurs when the
piezoelectric material becomes strained when placed in an electric field. The ability to
induce strain can be used to generate a movement, force, pressure, or vibration through
the application of a suitable electric field. The most promising commercial piezoelectric
materials are lead zirconate titanate (PZT) and polyvinylidene fluoride (PVDF).
PZT’s offer low strains (~0.06%) with significant hysteresis (~15-20%), where as
electrostrictive PMN materials exhibit higher strains (~0.1%) with lower hysteresis (~1-
4%). However, the temperature operating limits for PMN would require that it be
specially insulated. For all other considerations, PMN would be handled similarly to
PZT and actuator designs for one material could be applied to the other. While PZT is a
preferred material for most applications, future commercial applications may favor a
direct replacement of PZT with PMN because of its increased strain capabilities and
superior hysteresis efficiency. However, piezoelectrics have high bandwidths, they are
more compact than magnetostrictive devices and they are bi-directional, unlike
electrostrictives.
The potential now exists for electric actuators to deliver high forces over large
displacements. Larger displacements can be achieved by stacking piezoelectric or
electrostrictive elements. These actuators are being used in micro-positioning xy-tables,
3
ultrasonic motors, impact printer heads, vehicle suspensions, and precision machining
equipment [2-5].
The application of electric field controlled actuators has also been addressed in a variety
of space structures applications. Primary attention has been given to research in the area
of variable geometry truss structures [6-10]. For example, electrostrictive actuators are
used in the correction of the aberrations in the Hubble Telescope. Variable geometry
truss structures can adopt a wide range of geometric configurations by lengthening or
shortening some of their active elements, each containing a strain-induced actuator. This
ability allows these structures to adopt configurations that maximize their structural
strength. These structures may also use strain-induced actuators to produce damping
effects to suppress vibrations [11,12]. This technology has been integral to many space
truss systems including large span roof trusses, space reflectors, and robotic arms [6].
Other applications include release mechanisms, positioning devices, shape control of
large flexible surfaces, dexterity and obstacle avoidance [13].
Past Inchworm Designs 1.1.
During the last 40 years of development, piezoceramic inchworm actuators have
concentrated around a common theme. This theme incorporates piezoceramic stacks for
gripping and extending which create the actuator motion. In 1964, Stibitz [14] used a
magnetostrictive material on the end of three rods. Each of the three rods would grip,
release or push the inner shaft in a predetermined sequence that would actuate the shaft.
McNancy [15] developed an amplification device consisting of a piezoceramic stack
that was specially positioned against ball bearings. The displacement of the
piezoceramic was transferred and amplified though the ball bearings. In 1966, Hsu [16]
4
designed an actuator using a hollow cylinder with an inner clamping and extending
device.
In 1967, Locher [17] developed a mechanism that used two cams to grip and release an
inner shaft. The center shaft contained a piezoceramic material producing the extension.
In 1968, Brisbane [18] invented an inchworm actuator using a tube and an inner
crawler. The crawler had three piezoceramic elements: two for gripping and one for
extension. In 1972, Galutva [19] designed an actuator, which consisted of several
piezoceramic elements used for gripping and extending.
In 1975, Bizzigotti and May [20] introduced an inchworm actuator that used curved
surfaces to grip the outside of a shaft. The piezoceramic on one end would grip the shaft
while the center piezoceramic would extend. After extension, the piezoceramic on the
other end would grip the shaft to capture the displacement. In 1976, Sakitani [21]
invented the inchworm actuator that would clamp and extend on a surface providing a
precise displacement.
In 1979, Ishikawa [22] developed an actuator incorporating the use of two extending
and two clamping piezoceramic elements. When energized in a particular sequence the
upper shaft would be forced to move horizontally. In 1980, O’Neill [23] presented the
first stacked piezoceramic actuator. The idea of stacking increased the actuator
displacement. The design involved a cylindrical tube with an inside crawler. In 1984,
Taniguchi [24] presented an actuator that used an outer cylindrical shell and an inner
crawler. The inner crawler consisted of several cylindrical piezoceramic elements
connected together. The firing sequence of the cylindrical piezoceramics was such that a
“rippling” motion of extending and clamping made the inner body move. In 1986, Hara
5
[25] modified the design of Bizzigotti and May by incorporating stacked piezoceramics
to increase the actuation displacement. In 1986, Staufenberg [26] created an actuator
capable of translational and rotational motion. Many piezoceramics were used in the
design. Some were used to grip the shaft were others were used to push the shaft
outward (translational motion), or to push the shaft sideways (rotational motion). In
1988, Fujimoto [27] filed a patent consisting of two similar inchworm actuators that use
an inner crawler to creep inside a channel.
In 1990, Murata [28] used one set of piezoceramics to engage a shaft with an extremely
small pitch. Once engaged, the shaft would be actuated by another set of piezoceramics.
In 1991, Shibuta et al. [29] designed an actuator for precise pointing and vibration
suppression for anticipated use on the Geostationary Platform planned by NASDA. In
1994, Rennex [30] used piezoceramics with flexure clamps to hold and actuate the inner
shaft. In 1996, Pandell and Garcia [31] designed an actuator similar to the design of
Galutva but having one extra stage for extension and clamping. In 1997, Galante [32]
designed an actuator having an inner shaft that would move with respect to the outer
casing. The inner shaft had one piezoceramic used for extension, where as the outer
casing had two piezoceramics used for gripping. Firing the piezoceramics in a special
sequence forced the inner shaft to move. In 1999, Canfield et al. [33] developed an
actuator for minimally invasive surgery. The proposed mechanism consists of an
inchworm that moves a set of jaws (machined from titanium by the wire EDM process).
When the inchworm moves back and forth the jaws are force to clamp and unclamp.
Also in 1999, Frank et al. [34] developed an inchworm for flow control.
6
Unfortunately, many of these designs rely on extremely high tolerances (~2.54µm) with
no method of adjustment. In the case of Frank et al. [34], for example, the clearance
between output shaft and the gripping surface is less than 5µm. This exceptionally high
tolerance makes the actuator extremely delicate. Any tiny imperfections in machining
results in significant loss of performance or prevent actuator motion entirely.
Furthermore, most of the inchworm actuators are patented ideas and there has not been
an effort to study in detail the design to provide real proof-of-concept performance data.
The only designs that have reported performance results are the actuators proposed by
Pandell and Garcia with an output force of 13N and a static holding force of 44N with a
maximum speed of 1mm/s. The gripping device proposed by Canfield et al. produced a
block force of 44N and also a maximum speed of 1mm/s at a frequency of 100Hz.
This paper proposes an improved design and proof-of-concept for an electrostrictive
inchworm actuator. The design is based on the actuator’s ability to maintain a locked
position with no electrical power, fewer number of parts to reduce manufacturing
complexity, and its adjustability. The design goals of the actuator are to meet the
performance specifications presented in Table 1.
2. ACTUATOR DESIGN AND MODELING
The inchworm actuator concept uses small incremental steps to attain large
displacements. This motion is achieved by a mechanism that “crawls” inside an outer
casing, as illustrated in Fig. 1. The walking mechanism consists of two flextensional
brake assemblies separated by a center electrostrictive stack. Each brake assembly is
forced to clamp and unclamp in a particular sequence to capture the displacement of the
7
center stack. Consequently, the brake assemblies are the most critical aspect of the
actuator design.
The Actuator Stacks 2.1.
The amount of strain produced in a piezoceramic (piezoelectric or electrostrictive
material) is dependent on the thickness of the element and the magnitude of the voltage
applied across the thickness. To increase the amount of displacement, many thin
piezoceramic elements can be stacked on top of each other. In this stack arrangement,
the total stack displacement is equal to the number of elements multiplied by the
displacement of each element:
tnEd 3333 =δ (1)
where 3δ is the zero load deflection, is the piezoelectric charge coefficient, 33d E is
the electric field, t is the thickness of each stack wafer, and is the number of wafers n
The stack is made of electrostrictive material manufactured by Sensor Technology Ltd.
[35]. Construction of a stack is achieved by gluing many “wafers” of active material on
top of one other. Before the wafers are ready for stack assembly, they must be
individually fired, silvered, and poled. The wafers must be protected from excessively
high temperatures that will cause the active material to de-pole. The stacks are then
encased in a protective wrapping. Lead wires protrude from the encasing for electrical
connection. Metal washers are placed at each end to evenly distribute the load to the
brittle ceramic. Loads must be compressive and perfectly centered to not damage the
stacks.
8
Using the manufacturer’s suggested voltage of 200V with a stack wafer thickness of
0.508mm, the electrostrictive material yields an electric field of 0.3937MV/m. The
corresponding strain is 0.0567% (or 2.88x10-5mm per stack element). Using these
values, the calculated d33 is equal to 0.000567/0.3937x106 = 1440pC/N. The extending
stack of the inchworm has 133 layers (0.381mm each) and a total height of 76.2 mm.
This stack provides a displacement of 39.9µm when subjected to 200V. table 2 presents
a comparative analysis of the free displacement between the pieozlectric and
electrostrictive stacks. As observed, the free displacement for the electrostrictive stacks
is approximately four times greater than the piezoelectric stacks.
2.2 The Flextensional Frame
The flextensional frames are designed to have an interference fit with the outer casing.
When the brake stacks are energized, the frames are forced to distort and reduce in
width. This action frees the flextensional frames and allows them to move freely within
the outer casing. When the stacks are de-energized, the frames grip the outer casing
locking the actuator in place.
Firing each stack in a predetermined sequence creates the actuator motion. First, the left
piezo stack is energized. This action compresses the pre-stress material and forces the
diagonal arms connected to the brake pads to move up and in; this action releases the
top brake pads. Next, the center stack is actuated, pushing the top brake assembly
upward. Afterward, the left stack is de-energized, forcing the brake pads to grip the
outer casing. The right stack is then fired, releasing the pads. Subsequently, the center
stack is de-energized, moving the right brake assembly up. Finally, the right stack is
9
de-energized, forcing the pads to grip the outer casing and capture the displacement.
This sequence is repeated in rapid succession emulating the movement of an inchworm.
The material for the flextensional frame was chosen to have a coefficient of thermal
expansion similar to that of the electrostrictive stack to minimize the thermal effect so
as to reduce any temperature dependant dimensional variation between the stack and the
surrounding material.
The metals chosen must have a high Young’s modulus to efficiently transfer the stack
force to the brake pads, and a coefficient of thermal expansion similar to the selected
actuator stack material. The materials with the highest Young’s modulus are oil
hardened tool steel, stainless steel 304, and titanium (6% Al, 4%V). The titanium was
chosen, however, because it is the closest match to the coefficient of thermal expansion
of the piezoceramic (~1x10-6/C). Moreover, titanium (6%Al, 4%V) has a high strength
to weight ratio (density = 4730Kg/m3).
A rod with adjusting nuts has been inserted through the center of the actuator. In this
design, each brake assembly incorporates two pre-stressing adjustments: one for the
stack and one for the frame. As shown in Figure 2(a), when adjustment nuts 1 and 2 are
tightened, the stack is pre-stressed. This ensures the stacks always remain in
compression. When adjustment nuts 1 and 3 are tightened, the frame is pre-stressed.
Pre-stressing the frame allows a fine width adjustment within the outer casing.
10
2.3 Parametric Optimization
The dimensions were broken down into two categories: the fixed dimensions and the
free dimensions.
Fixed Dimensions
The fixed dimensions were set due to the size and shape of the electrostrictive stacks
quoted from the manufacturer. These stacks have an outer diameter of 25.4mm, an inner
diameter of 6.35mm, and height of 57.15mm. The resulting fixed dimensions are shown
in Fig. 2(b).
The minimum clearance required to adjust Nut 2 was 10mm. The pre-stress on the PMN
stack was chosen to be 4.45N/mm2 as this is significantly lower than 40N/mm2, which
is the maximum compressive loading for typical piezoceramic stacks [36]. The last
fixed dimension was the threaded rod diameter. It was chosen to be 4.7625mm since it
was the next nominal size smaller than the PMN stack hole of 6.35mm.
Free Dimensions
The free dimensions are presented in Figure 2(c) The dimensions having the greatest
impact on the design were selected first. The order in which the dimensions were
examined is the following: notch thickness (NT). shoulder thickness (ST), arm thickness
(AT), arm angle (AA), notch diameter (ND), and casing pre-stress (CP).
2.2. Performance Criteria
To assist in the selection of the free dimensions, three performance criteria were
defined: the range, the block force and the fatigue safety factor, as outlined next.
The Range
11
Referring to Fig. 3(a), when adjustment nut 3 is tightened, the flextensional frame bows
outward and the brake pad moves out to position “A”. Now energizing the PMN stack
forces the threaded rod to lengthen allowing the flextensional frame to move inward to
position “B”. The distance between brake pad locations “A” and “B” is defined as the
“range”. The range is important because it is a measure of the brake pad movement
during actuation. Too small a range may not allow the brake pad to free itself from the
outer casing.
The Blocked Force
The brake pad blocked force is related to the actuator output force through the
coefficient of friction constant “ µ ” defined by NFout **2 µ= , where is the
actuator output force,
outF
µ is the static coefficient of friction, and is the brake pad
blocked force. The coefficient of friction constant “
N
µ ” is dependant on many factors
such as material, counter-material, lubrication, temperature, speed, loading force,
surface finish, surface finish of the counter-material, and type of motion (reciprocating,
rotating). Published values can have large variations and range from 0.05 to 0.90. The
inside crawler is made from titanium and the outside casing can either be made from
titanium or oil hardened tool steel. A typical value is 0.65 for titanium against steel and
0.38 for titanium against titanium [37]. Using these values the actuator output forces are
presented in Table 3. For reduced cost of machining and increased frictional force the
material choice for the prototype is oil hardened tool steel.
For the brake pad to come in and out of contact, the outer casing must be located
somewhere between the maximum and minimum brake pad lateral displacements. To
12
ensure the brake pads release from the outer casing during actuation there should be a
clearance. The minimum suggested clearance is 25.4µm. To make certain the pad
release, the design aims for a clearance between 25.4µm and 50.8µm. For the brake pad
to come in and out of contact, the outer casing (“C”) must be located somewhere
between “A” and “B”, as shown in Fig. 3(b).
Suppose the PMN stack was energized, and the location of the outer casing “C” was
positioned at “B”. The brake pad and the outer casing would be in contact but without
force. If the PMN stack was now de-energized, the flextensional frame would try to flex
back to position “A” but would be stopped by the outer casing. The normal force
exerted on the outer casing is coined the “zero clearance blocked force”.
The Fatigue Safety Factor
The operation of the actuator requires the flextensional frame to bend back and forth a
high number of cycles. Each cycle induces a fluctuating stress, which over time may
results in the possibility of fatigue failure. To safeguard against this, a fatigue analysis
was preformed. Using the modified Goodman relation [38], along with the surface
roughness taken as machined, the allowable stress amplitude was calculated.
Subsequently, the actual stress amplitude was also determined for each design. Dividing
the allowable stress amplitude by the actual stress amplitude gives the fatigue safety
factor. Investigating the possible locations for fatigue failure it was determined that the
most critical location was at the center notches.
13
The maximum and minimum Von Mises stresses were analyzed. Using these upper and
lower limits of stress, a fatigue analysis was preformed. The modified Goodman
relation rearranged for aσ is presented in Eq. (2),
1=+ut
m
e
a
SSσσ
(2)
where aσ is the stress amplitude, is the endurance limit, eS mσ is the mean stress,
is the ultimate stress and utS
)1(*ut
mea SS
σσ −= (3)
The minimum ( minσ ) and maximum ( maxσ ) stresses were evaluated. The mean stress
( mσ ) was then calculated using Eq. (3),
2
)( minmax σσσ
+=m (4)
Using this value of mean stress ( mσ ) in Equation (3) with the endurance limit of
machined titanium ( 279MPa) and the ultimate strength of titanium ( =900MPa)
yields the maximum allowable stress amplitude (
=eS utS
allowa σσ = ). Now, using the minimum
( minσ ) and maximum ( maxσ ) stresses the actual stress amplitude ( actσ ) is determined
using:
2
minmax σσσ
−=act (5)
14
Finally the “Fatigue Safety Factor” is calculated by dividing the allowable stress
amplitude by the actual stress amplitude as:
act
allowFSσ
σ=.. (7)
Investigating all the possible locations for fatigue failure it was determined that the most
critical location was at the center notches of the flextensional frame.
2.3. Results
The parametric analysis results are quite extensive. For illustrative purposes, some
representative graphs and trends are shown in Figures 4 and 5.
Notch Thickness (NT): the first and most sensitive dimension that was studied was the
notch thickness. To study the notch thickness all dimensions were fixed (to an initial
guess at the best design) while only the notch thickness was varied. The notch thickness
was varied from 0.75mm to 2mm. The design goal of the brake assembly aims for a
range of not less than 90µm, a blocked force of over 15N (on each brake pad), and a
fatigue safety factor greater than 2.5. A range of 90µm should provide adequate
adjustment should there be any slight tolerance inconsistency when the design is
prototyped. A minimum force of 15N on each brake pad will generate a 30N normal
force on the outer casing. This force multiplied by the coefficient of friction for the
outer casing material (usteel on Ti = 0.69) will lead to a acceptable actuator pushing force
of 20.7N. A fatigue factor of safety over 2.5 will safeguard against the high number of
cycles the design must endure over its lifetime. Choosing a notch thickness of 1.00mm
gives acceptable values for the range, the blocked force, and the fatigue safety factor.
15
Shoulder Thickness: The next dimension that was considered was the shoulder
thickness. Again, all the dimensions were fixed except the shoulder thickness which
was varied from 5mm to 10mm. Increasing the shoulder thickness favorably increases
both the range and the blocked force; unfortunately, the weight, and the actuator length
also increase. A tradeoff between these parameters was met by selecting a shoulder
thickness of 9mm.
Arm Thickness: The thickness of the arms was examined next. The arm thickness was
varied from 4mm to 8mm. Increasing the thickness of the arms increases range and
blocked force, but decreases the fatigue safety factor. A compromise between these
variables was chosen by selecting the arm thickness to be 5mm.
Arm Angle: The next free variable that was studied was the arm angle (AA). The angle
was varied from 1.19° to 9.42°. Interestingly, as the angle of the arm increases past
3.49°, the range begins to decline. Analyzing, the points before and after AA=3.49°
shows the arms undergo a “snap through” condition. This means for angles of AA less
than 3.49° (AA < 3.49°) the arms are concave where as at angles above 3.49° (AA >
3.49°) the arms are convex. If the arm angle was chosen to be 3.49° the range would be
at its maximum of 109µm with the a resulting blocked force of 28N (clearance=50.8N)
and the fatigue safety factor of 2.9. However, if the arm angle was chosen to be 4.87°
the range would be decreased slightly to 106µm along with a decrease in the fatigue
safety factor to 2.8, but the blocked force would be drastically increased to 39N.
Therefore, the angle of the arm was chosen to be 4.87°.
16
Notch Diameter: The next free variable that was tested was the notch diameter. The
notch diameter was varied from 1mm to 3mm. The notch diameter has little effect on
the range or blocked force. Therefore, the notch thickness was chosen to be 2.5mm to
ensure a acceptable value of 3.3 for the fatigue safety factor.
Casing Pre-Stress: the last free variable that was investigated was the casing pre-stress.
The casing pre-stress was varied from 400N to 800N. The range, blocked force, and
fatigue safety factor decrease as the casing pre-stress is increased. Therefore, it is
desirable to have a low casing pre-stress. However, the casing pre-stress must not be so
low that Nut 3 comes out of contact with the outer casing during actuation. Choosing a
casing pre-stress of 565N provides a force of 206N between Nut 3 and the casing during
actuation. Note that for adjustment Nut 1 the positive force is taken downward where as
for adjustment Nuts 2 and 3 the positive force is taken upward. The dimensions of the
brake assembly for the final design are shown in Fig. 6.
3.
3.1.
ACTUATOR PROTOTYPE AND TESTING
A prototype of the computer model was built to attain physical measurements to
validate the final design of the actuator. To accomplish this, the procedure was
performed in three steps: the assembly of the inner crawler, the manufacturing of the
outer casing and the design of the controller. Upon completion of each component of
the actuator, the prototype was assembled and tested.
The Prototype
A pictorial representation of the stacks can be seen in Fig. 7. Two brake stacks and one
extending stack were used to assemble the inner crawler.
17
The inner crawler is assembled from two flextensional frames, two brake stacks, one
center stack, and a center rod with adjustment nuts to hold everything together, as
shown in Fig. 8. Because the flextensional frames have a complex geometry and need to
be manufactured with a tolerance of 12.7µm , Electro Discharge Machining (EDM) was
the chosen manufacturing method. This process cuts each solid block of titanium using
a fine copper wire passing a high voltage enabling tolerances up to 2.54µm.
It should be noted that the prototype frame has extra protrusions (called, guide bumps)
along its outer perimeter. These protrusions were an after thought to help keep the inner
crawler aligned within the channel during operation. Since the guide bumps are located
in regions of low stress, they will not cause any significant discrepancies with the
computer model results.
After the flextensional frames were machined, the inside crawler was ready for
assembly. To accomplish this, the components were fitted onto the threaded rod and
pre-stressed using the adjustment nuts. In order to apply the proper pre-stress the nuts
are tightened using a two step sequence.
Initially, Nut 3 is loose and the stack is not energized. Choosing a pre-stress of 565N,
Nut 1 and Nut 2 are tightened to a value of 2676N, while Nut 3 remains loose. Now,
Nut 1 is tightened against Nut 3 to a value of 565N. The action of tightening Nut 1 and
Nut 3 reduces the force between Nut 1 and Nut 2 to a value of 2111N. The
corresponding pressure on the stack is therefore 4.45N/mm2, (which is the
recommended pre-stress by the manufacturer [35]).
18
The nuts on the center stack are also tightened to a value of 2111N to achieve a stack
pre-stress of 4.45N/mm2. To apply the correct pre-stress the necessary tightening torque
was evaluated using:
2sectan1
sectan2
np
r
rr dFFdT
µαλµαµλ
+
−
+= (7)
where T is the tightening torque, is the pre-stress force, d is the mean diameter of
threaded rod,
F r
rµ is the coefficient of friction on nut threads, λ is the thread lead angle,
α is the thread angle, pµ is the coefficient of friction of nut surface against the stack,
and is the outer mean diameter of nut. nd
The rod is made from titanium with a 10-32 thread ( = 4.318x10rd-3m, λ = 3°,
α = 30°). The outer mean diameter of the nut is = 2.395 x10nd-2m. The coefficient of
friction is taken as a lubricated thread ( rµ = pµ = 0.15). Using this information to
achieve a pre-stress of 2676N the necessary tightening torque is 6.12N·m. Similarly, to
achieve a pre-stress of 565N the necessary tightening torque is 1.29N·m.
Insertion of the inner crawler into the outer casing is shown in Fig. 9. To create motion
of the inner crawler, the stacks need to be fired in a particular sequence. This is done by
means of a controller.
3.2. Controller Design
A controller was designed to control the actuation of each stack. The controller was
made with two modes: one for assembly, and one for actuating. The assembly mode,
energized the braking stacks to allow insertion of the crawler into the outer casing. The
19
actuation mode was designed to fire each stack in a sequence that would cause the inner
mechanism to linearly translate.
In actuating mode the waveforms supplied to the stacks were varied in frequency,
amplitude, duty cycle and form. This was accomplished by using a PicStic 4x
microprocessor that was programmed in the PicBasic language. The features of the
microcontroller consist of: its two channel 12 bit digital to analog converter, its 8
bi-directional bit programmable high current input output lines and its high performance
processor. In order to supply a voltage differential to the stacks to excite them a
two-step power amplification circuit was designed. The first stage amplified the
waveforms output by the PicStic using high accuracy instrumentation amplifiers. The
amplifier design allows a very high bandwidth over a wide range of gains making it
ideal for rapid data acquisition [39]. The amplifier gain is set using a single external
resistor (RG) and can range from 1 – 10,000. The gain, G, is given by
150+
Ω=
GRkG (8)
For a chosen value of 33kΩ the gain is 2.5. The output from the amplifier can then
be further amplified by using a high voltage power amplifier.
GR
Three model SA-10 high voltage power amplifiers and one SA-20 high voltage power
supply were obtained from Sensor Technology Limited. The SA-10 is a two-channel
high voltage power amplifier, it can be used as two individual ground referenced
amplifiers providing a gain of 15, or a single bridged amplifier providing a gain of 30.
Each channel gives a maximum ground referenced voltage variation of nominally 280 V
20
peak-to-peak. In bridged mode, the SA-10 provides a differential voltage variation of
560V [40]. During prototype testing the high voltage amplifiers and instrumentation
amplifiers were configured to obtain a voltage of 285V across each stack. The diodes
are used as a visual representation of the firing of each stacks, where R1 is 333Ω. A
picture of the circuit is shown in Fig. 10.
The actuating mode of the controller has two options. The first used square waves
output from the PicStic that were then varied in amplitude, frequency and duty cycle.
These waveforms were constructed using high and low logic levels cycled over chosen
input-output pins of the PicStic. This firing sequence is intended to make the inner
crawler move within the channel with a maximum controller frequency of 170Hz. The
second actuation option uses a ramping waveform applied to the center stack.
From previous papers [33,34] it was discovered that ramping the signal of the center
stack provided an increase in velocity of similar actuators when compared against the
square wave signal at frequencies above 100Hz (due to less jarring motion). The
ramping of the center stack was accomplished by using the digital to analog converter of
the input-output coprocessor on the PicStic; to create four intermediate steps between
the high and low voltage levels. Because the ramping function requires several
intermediate steps, the maximum operating frequency of this signal is limited to 45Hz.
3.3. Prototype Results
The prototype was tested. The experimental and numerical results are presented in
Table 4.
21
The quickest time the stacks can become charged (87% of maximum charge) or de-
charged (13% of maximum charge) is 0.075s. Since there are six steps in each cycle of
the actuator motion, the quickest time for one cycle of the inchworm movement is
6*0.075s = 0.45s. This corresponds to a maximum operating frequency of 2.22Hz.
Multiplying this value by the extending stack displacement per step (7.2µm) gives the
theoretical actuator speed of 1.0mm/min. The difference in results between the model
and the prototype differ by less than 9%, therefore validating the modeling and proof-
of-concept demonstrator.
4. CONCLUSIONS
This paper investigates the design of an electrostrictive inchworm actuator. The design
is based on its ability to maintain a locked position with no electrical power, fewer
number of parts to reduce manufacturing complexity, and its adjustability. Parametric
optimization design analysis of the brake assembly model were performed to select
dimensions providing appropriate force/displacement characteristics.
To validate the computer model, a prototype was built. Frequency response (and
actuator output speed) was limited in this case by the stack RC constant, however
improvement in the stack design and manufacturing would increase the response speed
of the actuator. The controller driving circuit has a maximum upper limit of 170Hz,
whereas the limiting frequency of the inchworm cycle is 2.22Hz. Future work may be
done to the stacks to enhance the maximum operational frequency and displacement.
Moreover, the natural frequency of the actuator system may be investigated through
future experimentation. At high frequencies, the ramping signal of the controller may
also prove useful.
22
5. REFERENCES [1] K. Uchino, “Materials Issues in Design and Performance of Piezoelectric
Actuators- An Overview”, Acta Materials, Vol. 46, No. 11, pp3745-3753 , 1998.
[2] K. L. Koudela and B. Hassier, “A 500Hz Mechanical Amplifier for use with
Multilayered Piezoelectric Actuators,” Proceedings of The Ninth International
Conference on Adaptive Structures and Technologies, Boston, Massachusetts,
U.S.A., p13-21, October 1998.
[3] G. R. Lui, Z. Yili, K. Y. Lam, X. Q. Peng and J.Tani, “Finite Element Modelling
of Piezoelectric Sensors and Actuators Bonded in Thick Composite Laminates,”
Proceedings of The Eighth International Conference on Adaptive Structures and
Technologies, Wakayama, Japan, p113-122, October 1997.
[4] D. R. Martinez, T. D. Hinnerichs and J. M. Redmond, “Vibration Control For
Precision Manufacturing Using Piezoelectric Actuators,” Proceedings of The
Sixth International Conference on Adaptive Structures and Technologies, Rome,
Italy, p3-22, October 1997.
[5] B. Pletner and H. Abramovich, “Piezoelectric Sensors For Adaptive
Suspensions,” Proceedings of The Sixth International Conference on Adaptive
Structures and Technologies, Rome, Italy, p404-417, October 1997.
[6] Y. B. Cohen, S. Leary, M. Shahinpoor, J. O. Harrison and J. Smith, “Electo-
Active Polymer (EAP) Actuators for Planetary Applications,” Proceedings of
The SPIE Smart Structures and Materials, Newport Beach, California, Vol.
3669, p57-63, March, 1999.
23
[7] P. C. Hughes, W. G. Sincarsin and K. A. Carroll, “Trussarm-A Variable
Geometry Truss Manipulator,” Proceedings of The First Joint Japan/US
Conference on Adaptive Structures, Nagoya, Japan, p715-725, November 1991.
[8] S. Shao and Y. Murotsu, “Some Approaches to the Optimal Adaptive
Geometries of Intelligent Truss Structures,” Proceedings of The First Joint
Japan/US Conference on Adaptive Structures, Nagoya, Japan, p743-771,
November 1991.
[9] K. Uchino, “Recent Development of Piezoelectric Actuators for Adaptive
Structures,” Proceedings of The Third International Conference on Adaptive
Structures and Technologies, San Diego, California, U.S.A., p245-257,
November 1992.
[10] A. Benjeddou, M. A. Trindade and A. Ohayon, “A Finite Element Model for
Shear Actuated Adaptive Structures,” Proceedings of The Eighth International
Conference on Adaptive Structures and Technologies, Wakayama, Japan, p133-
142, October 1997.
[11] M. E. Regelbrugge, and B. J. Huribut, “Developing of a Self-Sensing Multilayer
Piezoceramic Actuator for Structural Damping Applications,” Proceedings of
The Fourth International Conference on Adaptive Structures and Technologies,
Cologne, Federal Republic of Germany, p29-42, November 1993.
[12] H. Janocha, D. J. Jendritza, and J. Schafer, “Active Damping of Forced Periodic
Vibrations with Solid State Actuators,” Proceedings of The Fourth International
24
Conference on Adaptive Structures and Technologies, Cologne, Federal
Republic of Germany, p497-511, November 1993.
[13] M. T. D’Eleuterio and G. B. Sincarsin, “Articulational Dynamics of Variable-
Geometry Truss Structures”, Proceedings of the Second Joint Japan/U.S.
Conference on Adaptive Structures, Nov 12-14 1991, 1992, Nagoya, Japan, p
621.
[14] R. Stibitz, “Incremental Feed Mechanisms”, U.S. Patent: 3,138,749, 1964.
[15] M. McNancy, “Inchworin actuator”, U.S. Patent: 5,3323,942, 1994.
[16] K. Hsu and A. Biatter, “Transducer”, U.S. Patent: 3,292,019, 1966.
[17] G.L. Locher, “Micrometric linear actuator”, U.S. Patent: 3,296,467, 1967.
[18] A.D. Brisbane, “Position control device”, U.S. Patent: 3,377,489, 1968.
[19] G. V. Galutva, “Device for precision displacement of a solid body”, 1987.
[20] R. A. Bizzigotti, “Electromechanical translational apparatus”. US. Patent:
3,902,085, 1975.
[21] Y. Sakitani, “Stepwise fine adjustment”, U.S. Patent: 3,952,215, 1976.
[22] M. Ishikawa and Y. Sakitani, “Two-directional piezoelectric driven fine
adjustment device”, U.S. Patent: 4,163,168, 1979.
[23] G. O'Neill, “Electromotive actuator”, U.S. Patent: 4,219,755, 1980.
25
[24] T. Taniguchi, “Piezoelectric driving apparatus”, US Patent: 4,454,441, 1984.
[25] A. Hara, H. Takao, Y. Kunio, T. Sadayuki and N. Keiji, “Electromechanical
translation device comprising an electrostrictive drive of a stacked ceramic
capacitor type”, U.S. Patent: 4,570,096, 1986.
[26] C. W. Staufenberg and R. J. Hubbell, “Piezoelectric electromechanical
translation apparatus”, U.S. Patent: 4,622,483, 1986.
[27] T. Fujimoto, “Piezo-electric actuator and stepping device using same”, U.S.
Patent: 4,714,855, 1987.
[28] T. Murata, “Drive apparatus and motor unit using the same”, U.S. Patent:
4,974,077, 1990.
[29] S. Shibuta, Y. Morino, Y. Shibayama and K. Sekine, “Adaptive Control of
Space Truss Structures by Piezoelectric Actuator”, Tsukuba Space Cent Source:
Jt Japan/US Conference on Adaptive Structures, Proceedings of the Second Joint
Japan/U.S. Conference on Adaptive Structures, Nov 12-14 1991, 1992, Nagoya,
Japan Sponsored by: PA, Publ by Technomic Publ Co Inc, Lancaster, USA, p
245.
[30] G. Rennex, “Inchworm actuator”, U.S. Patent: 5,3323,942, 1994.
[31] T. Pandell and E. Garcia, “Design of a piezoelectric caterpillar motor”
Proceedings of the ASME aerospace division, AD-Vol. 52, p. 627-648, 1996.
26
[32] T. P. Galante, “Design and Fabrication of a High Authority Linear Piezoceramic
Actuator: The PSU H3 Inchworm”, The Pennsylvania State University: Graduate
School, Department of Mechanical Engineering, August 1997.
[33] S. Canfield, S. Edinger, B. M. Frecker, and G. Koopmann, “Design of
Piezoelectric Inchworm Actuator and Compliant End-Effect for Minimally
Invasive Surgery”, Society of Photo-Optical Instrumentation Engineers, p 835-
843, Part of the SPIE Conference on Smart Structures and Integrated Systems,
Newport Beach California, March 1999.
[34] J. Frank, G. Koopmann, W. Chen and G. A. Lesieutre, “Design and Performance
of a High Force Piezoelectric Inchworm Motor”, Proceedings of SPIE - The
International Society for Optical Engineering , v 3668, n II, Proceedings of the
1999 Smart Structures and Materials - Smart Structures and Integrated Systems,
1999, Society of Photo-Optical Instrumentation Engineers, Bellingham, WA,
USA, p 717-723.
[35] Sensor Technology Limited, www.sensortech.ca , 2001.
[36] V. Giurgiutiu and C. A. Rogers, “Energy-Based Comparison of High-Power
Commercially-Available Induced-Strain-State Actuators,” Proceedings of The
Sixth International Conference on Adaptive Structures and Technologies,
Rome, Italy, p113-130, October 1997.
[37] J.E. Shigley and C.R. Mishcke, Mechanical Engineering Design, 5th Edition,
McGraw-Hill Book. Co. 1989.
27
[38] Vishay Measurements Group, Inc., 2001,
http://www.measurementsgroup.com/guide/glossary/vonmises.htm, (Aug. 2001).
[39] Analog Devices, “High Accuracy 8-pin Instrumentation Amplifier – AMP02”,
Rev. D, Norwood MA USA.
[40] Sensor Technology Limited “SA-10 High-Voltage Power Amplifier Technical
Manual”, Rev. 2, March 23, 2000.
28
Table 1 - Actuator design specifications
Minimum Displacement (mm) 15Minimum Force Output (N) 20Maximum Operating Voltage (V) 200
Actuator Specifications
Table 2 - Comparison of free displacement of BM500 and BM600
Ansys Exp Ansys ExpFree Displacement (um) 7.2 7.3 30.0 28.8
PZT PMN
Table 3 Actuator output force
BrakePad
Clearance(um) Ti on Steel Ti on Ti
0 112 6525.4 84 4850.8 54 3276.2 25 15
ActuatorOutput Force
(N)
Table 4 - Summary of inchworm actuator results
Inchworm Ansys Ansys MeasuredActuator Design Prototype fromResults Model Model Prototype
Brake Stack Free Disp. (um)
30.0 27.1 26.0
Brake Pad Disp. (um)
107.4 78.2 71.1
Extending Stack Free Disp. (um)
39.9 16.7 16.0
Brake Pad Blocked Force (N): 25.4 Clearance 64.2 18.6 ---50.8 Clearance 41.6 7.5
Actuator Output Force (N): 25.4 Clearance 83.5 24.2 15.150.8 Clearance 54.0 10.0
Max. Operating Freq. (Hz) --- 2.22 0.33Actuator Speed (mm/min) --- 1.0 0.2
Figure 1 – The proposed electrostrictive inchworm actuator concept
(a)
(b) (c)
Figure 2 – (a) Schematic of the flextensional frame and stack with pre-tension nuts; (b) The
actuator fixed dimensions; and (c) the actuator free dimensions
(a) (b)
Figure 3- (a) The brake assembly range; and (b) the contact with outer casing
Notch Thickness Vs. Range(NT=Variable, ST=10, AT=10, AA=3°, ND=2.4, CP=650)
0.500.751.001.251.501.752.002.25
30 40 50 60 70 80 90 100 110 120 130 140
Range (um)N
otch
Thi
ckne
ss
(mm
)
Shoulder Thickness Vs. Blocked Force(NT=1, ST=Variable, AT=10, AA=3°, ND=2.4, CP=650)
4.505.506.507.508.509.50
10.50
0 10 20 30 40 50
Blocked Force (N)
Shou
lder
Th
ickn
ess
(mm
) Clearance=0.00um
Clearance=25.4um
Clearance=50.8um
Clearance=76.2um
Arm Thickness Vs. Max Von Mises Stress(NT=1, ST=9, AT=Variable, AA=3°, ND=2.4, CP=650)
3.504.505.506.507.508.50
100 150 200 250 300 350Von Mises Max Stress @ Center
Notches (MPa)
Arm
Thi
ckne
ss
(mm
)
Clearance=0.00um
Clearance=25.4um
Clearance=50.8um
Clearance=76.2um
During Actuation
Arm Angle Vs. Fatigue Safety Factor(NT=1, ST=9, AT=5, AA=Variable, ND=2.4, CP=650)
0.001.503.004.506.007.509.00
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5Safety Factor
Arm
Ang
le (d
eg)
Clearance=0.00umClearance=25.4umClearance=50.8umClearance=76.2um
Figure 4 – Parametric design optimization of free dimension design parameters
Notch Diameter Vs. Range(NT=1, ST=9, AT=5, AA=4.87°, ND=Variable, CP=650)
0.751.251.752.252.753.25
102 104 106 108Range (um)
Not
ch D
iam
eter
(mm
)
Casing Pre-Stress Vs. Range(NT=1, ST=9, AT=5, AA=4.87°, ND=2.5, CP=Variable)
350
450
550
650
750
850
103 104 105 106 107 108 109 110
Range (um)
Cas
ing
Pre-
Stre
ss(N
)
Figure 5 - Variation of notch diameter and casing pre-stress with free dimension design parameters
Notch Thickness (NT) = 1mm Shoulder Thickness(ST)= 9mm Arm Thickness(AT) = 5mm Arm Angle(AA) = 4.87° Notch Diameter(ND) = 2.5mm Casing Pre-Stress(CP) = 565N
Figure 6 - Final design modeling results
Figure 7 - Design and prototype stacks
Figure 8 – Flextensional frame and the inner crawler
Figure 9 - Assembly of outer casing and inner crawler
Figure 10 - Controller assembly and waveforms